【摘要】 作为新一代超宽禁带半导体材料，单晶金刚石（SCD）凭借其优异的物理特性（禁带宽度5.5 eV，击穿场强9.9 MV/cm，热导率22 W/(cm·K)），被视为突破硅基集成电路性能极限的理想材料。然而，其极高硬度（莫氏硬度10）和极强化学惰性使其表面加工面临重大挑战，特别是晶圆级平坦化技术已成为制约金刚石集成电路发展的关键瓶颈。聚焦集成电路制造需求，系统评述了单晶金刚石抛光技术的最新进展，重点分析了机械类、高能束和多场耦合三大类抛光方法的材料去除机理、影响因素及局限性。机械类抛光（如机械抛光和超声波辅助抛光等）工艺过程简单，但硬对硬摩擦易引入表面损伤，难以实现亚纳米级抛光；高能束抛光（如激光抛光、离子束抛光和等离子体抛光等）利用高能粒子代替磨粒来抛光，这类抛光存在选择性差、热影响区深、高能粒子注入等问题，难以实现低亚表面损伤及亚纳米级平坦化；多场耦合抛光（如化学机械抛光和等离子体辅助抛光等）通过场间耦合效应实现表面平坦化，但抛光工艺复杂且速率较低。尽管现有技术在不同应用场景中取得了一定进展，但仍无法完全满足亚纳米级精度、低损伤、高速率的抛光需求。未来的研究应致力于开发新型抛光技术，结合多种抛光方法的优势，推动单晶金刚石在高端芯片制造领域的广泛应用。更多还原

【Abstract】 Single-crystal diamond(SCD), owing to its exceptional physical properties—including an ultrawide bandgap of 5.5 eV, an extremely high breakdown electric field of 9.9 MV/cm, and an outstanding thermal conductivity of 22 W/(cm·K)—is widely recognized as a revolutionary material capable of overcoming the performance limitations of silicon-based integrated circuits(ICs). However, the extraordinary hardness(Mohs hardness of 10) and extreme chemical inertness of diamonds present significant challenges for achieving atomic-level surface polishing, which is crucial for their application in high-end chip manufacturing. This study focuses on the development of atomic-level polishing techniques for SCD, systematically reviews the evolution of polishing technologies from micro/nanoscale to atomic precision, and identifies key breakthroughs to overcome existing machining limitations. Additionally, this study examines the characteristics and applicability of various polishing methods, thus providing valuable insights for practical implementation. First a mechanical polishing techniques, including conventional and ultrasonic-assisted methods, are investigated comprehensively. Whereas these approaches offer straightforward processes, the inherent “hard-on-hard” friction inevitably introduces surface defects, thus rendering it difficult to achieve subnanometer surface smoothness. Subsequently, high-energy beam polishing technologies, such as lasers, ion beams, and plasma polishing, are examined. Although these methods replace abrasive particles with high-energy particles, issues such as inferior selectivity, deep thermally affected zones, and particle implantation limit their ability to achieve low subsurface damage and subnanometer planarization. The third category, i.e., multifield coupled polishing, which includes chemical-mechanical polishing and plasma-assisted polishing, leverages synergistic effects for surface planarization. However, these methods are characterized by complex processes and low polishing rates. Based on a detailed comparative analysis, this study highlights the significant challenges in satisfying the stringent requirements of IC manufacturing: subnanometer precision, minimal damage, and high processing rates. The findings suggest that, although existing polishing techniques are adopted in specific applications, they remain inadequate for completely satisfying the abovementioned demanding criteria. Achieving atomic-level surface polishing requires not only gradual process improvements but also systematic innovations in theoretical understanding and process development. This study emphasizes the necessity of cross-scale control from macroscopic process parameters to microscopic atomic behavior to precisely regulate material removal mechanisms, including the synergy between mechanical shear and chemical etching, as well as the energy threshold for atomic bond breaking. At the fundamental research level, the development of atomic-resolution in-situ characterization techniques and first-principles computational models is crucial for establishing quantitative relationships between process parameters and atomic surface configurations. The results indicate that atomic-scale manufacturing must rely on multifield synergistic regulation combined with in-situ atomic-level monitoring and intelligent control to achieve precise process optimization and further advance ultraprecision manufacturing. Despite significant progress, several technical bottlenecks remain in achieving atomic-level polishing for SCD. First, multifield coupling mechanisms are yet to be fully understood, thus resulting in trade-offs between the material-removal rate and surface quality in mechanical polishing, as well as issues such as high equipment costs and instability in energy-beam polishing. Additionally, multifield coupling techniques present challenges such as dynamic parameter mismatches(e.g., pH fluctuations and uneven light distributions), thus hindering stable and efficient processing. Second, intelligent control systems for polishing processes remain underdeveloped, with limited real-time optimization capabilities and insufficient integration of smart algorithms with in-situ characterization techniques. This results in a trial-and-error approach for process optimization. Third, the industrialization of green and efficient processes is hampered by key obstacles such as high energy consumption, environmental risks, and low process repeatability, which pose significant constraints for large-scale applications. Hence, future research should focus on three key directions:(1) deepening the understanding of multifield coupling mechanisms, including the interactions between mechanical forces, chemical etching, and energy fields(e.g., light, sound, and plasma), and establishing quantitative models of parameter synergy;(2) advancing intelligent control technologies, such as machine learning-based algorithms for real-time process optimization, and developing high-precision multifield coupling equipment to enhance process stability and consistency; and(3) promoting green and efficient processes, including the development of environmentally friendly chemical systems and energy-saving machining methods, as well as establishing standardized protocols for industrial implementation. The key innovation of this study is its systematic evaluation of polishing technologies and the identification of critical research directions to overcome diamond-machining challenges. By comprehensively assessing the strengths and limitations of existing methods, this study provides a solid foundation for the development of next-generation diamond-based ICs as well as offers valuable insights for academic researchers and industry professionals in advanced semiconductor manufacturing.更多还原